sign in sign up
<<
Home , Antirrhea, Caerois, Morphinae, Morpho, Morpho peleides, Opsiphanes

Rev. Biol. Trop., 23( 1): 10l-12�, 19n

Feeding behavior of (: : ) in a seasonal

tropical environment*

by

Allen M. YoungU

(Received for publication March 3, 1975)

ABSTRACT: Observations in a semi· tropical wet forest in Gua­ nacaste, Costa Rica, early in the dry season (December 1973) revealed that adults of the Morpbo pe/eides (Lepidoptera: Nymphalidae: Mor­ phinae) feed on tiny sap flows on exposed roots of the tree Samanea saman (Leguminoseae) . These observations supplement earlier studies on the association of the butterflies with fallen, decaying fruits of the tree Gllazllma II/mifolía (Sterculiaceae) in the same forest but during the previous dry season (January 1973). Morphoes collect in small groups of both sexes on the tree sap flows along with several of ants. These feeding clubs, which are formed daily, usually in the afternoons, are tightly struc­ tured in that adults feed very closely together with no indications of agonistic behavior or courtship. When suddenly disturbed, the butterflies cease feeding and fly around the area before settling on nearby vegetation. The feeding behavior is different when morphoes feed on juices of Gllazllma fruits : adults are wide1y dispersed and no feeding clubs are formed. GlIa zllma trees are more abundant in this forest and scattered over a larger area than the few Samanea trees. The association of lllnfl; with the tree sap flows and the conspicuous lack of Caligo memmon, both abundant butterflies, is discussed from the standpoint of phylogenetic relationship between the and Morphinae. The implications of the changeover in ieeding behavior oi Morpho are discussed in terms of seasonality and destruction of forests by '!l-gricultural practices. Experiments employing baits predict changes in the dispersion of the adult population at different times during the dry season.

'" This papee is dedicated to Dr. J. Robert Hunter, agronomist, educator, and long­ time resident of Costa Rica, for his services to the Costa Rican Field Studies Pro­ gram of the Associated Colleges of the Midwest. .. Department of Biology, Lawrence University, Appleton, Wisconsin 54911, U.S.A.

101 BiotOGIA TROpiCAL lói REVISTA bE

Behavioral responses of organisms can be very adaptive in that they provide the quickest means by which adjustments can be made to changes in environmental conditions affecting fitness, and if inadequate, they may eventually give way to more drastic response patteras such as physiological and genetic changes (SLOBODKIN, 8). Changes in behavior patteras in response to short­ term changes in the environment provide collectively a unique mechanism of adaptive flexibility. This mechanism provides a different sort of adaptive response than that originatíng in genetic changes since, depending on the duration of an environmental change, individual s in. a population may lag in a genetic adaptive response by one or more generations (LEWONTIN, 3) . . But behavioral responses can be extremely important forms of selection pressure during the early stages olf an organism entering into a new adaptive zone or new ecological niche (MAYR, 4). It is precisely for these reasons that studies of behavior in the wild are jmportant for understanding the mechanisms of adaptation in popula­ tions of selected species. Chariges in the behavioral response of a given animal to, say, food can be brought about by changes in the availability of a preferred of food. If local environmental conditions in the habitat result in a reduction in such a food source, the behavioral response may entail the switch to an alternative type df food which is more available at that time. If such a switch cannot be made, the animal is fated to either local extinction or forced emigration, which may be temporary or permanent. One might imagine that an animal living in a seasonal environment could be subject to the hecessity of switching food if the food availability fluctuates in a seasO!nal fashion. This study is conceraed with the adult feeding behavior of the neo­ tropical butterfly ¡impida (Lepidoptera: Nymphalidae : Mor­ phinae) in a seasonal lowland tropical forest in northwestein Costa Rica. An earlier repart cbncerning this particular popúlatiori of MorPho - (YOUNG & THO­ 13) MASON, suggested that conditions of environ-mental stress, imposed by a severe dry season, have_ pronounced affects ón demography of the species in this habitat. This preserit report discusses yet another aspect of this -stress : ' seasonal changes in food sources exploited by- adult Morpho result in a change in feeding behavior which very likely involves a switch from fruits to sap ' wounds on . . large fóre�t trees. _ As_ a grcitIp, the "morphoes" are among the luost familiar and spectacular of the ' butterfly fauna · in Central and South A-merica. Although most species are generally associated with rather - nonseasonal tropical wet forests (SElTZ, 7),

M. pe!eides ís known to inhabit very seasorial habitat� in El Salvador .al?:d Costa Rica (YOUNG & MUYSHONDT, 12). Along with several other genera of forest­ dwelling neotropical butterflies, mdstly in the Nymphalidae; the morphoes feed primarily on a variety of rotting, sweet-odoriferous fruits and fleshy fungus growths on the damp floor oE forests (7, 12), althbugh the behavior patterns associated with feeding have- not been �xplored. While this paper dea}s with

. obs�¡:vatiOl1s pE feeding behavior in adult Moypho, the impliéáticins , of these studies may be of general interest with respect to how tropical seasonality e affects the fe ding habits of . There is evidence that the Costa Rícan MORPHO YOl!NG; FEEDING BEHAVIOR OF tlUTTER.FLIES 103 dry season alters the kinds of active insects in the community (JANZEN & SCHOENER, 2); small insects become inactive to !lvoid desiccation. But even large, robust insects like the morphoes can be expected to show changes in behavioral responses related to feeding, as food sources change with the advan­ cíog dry season. There are several interesting applied implications of this study for tro­ pical entomolOlgy in general. First, there is the question of how seasonal climatic changes affect the feeding behavior of insects, and more important1y, how such changes affect insects that feed on plants or plant products. It is known that trees in the tropics are affected in various ways by seasonality (2), and seasonality may induce insects to changeover in foods as seasons change, and this may involve changes in nutritional intake in different seasons. This particular study has even other implications for applied studies since the insects in question were confined to a forest habitat isolated frem other similar habitats as a result of agricultural practices : it is a confined population of an ¡nsect that is dependent upon food items produced by tree species and not only have many of these trees been removed when the land was c1eared, but even today there is continued selective remc.cval of trees from the remnant forest itself. The Morpho butterflies present a very different problem of adaptation to man-made tropical habitats in that they depend upon food items other than flowers, which contrasts with Precis butterflies in A frica, where they feed on various flowers in gardens (OWEN, 6).

MATERIAL AND MEI'HODS

The site foe these studies, the "Barranca forest", in the trcrpical dry lowlands of northwestern Costa Rica is unique in that it is a small area of semi-deciduous troplcal forest (about 900 x 500 m) surrounded by open pasture (Fig. 1). A rather pronounced dry season usually begins in ear1y NOIVember and lasts through May, during which time many large trees lose their leaves and cease vegetative growth (5). A good description of this small forest is given by ORIANS (5). The forest supports a rich fauna of insectivorous reptiles and birds, and many of the nymphalid butterflies that are active near the forest floor are usuaUy wary when approached with a net, probably attribu­ table to severe predation pressure by these small vertebrates (YOUNG, 11). The butterfly fauna associated with rcrtting fruit and tree sap flows includes Prepona, HamadrYáf, Anaea, Historis, Victorina, Caligo, , and Satyrinae in addition to Morpho. The forest floor is . covered by a dense layer of leaf litter (Fig. 2) that renders most of these butterflies very cryptic while feeding on or near the floor. During a short period of eleven days in January of 1973, casual observa­ tions were made on the feeding behavior of Morpho on experimentally-placed

YOUNG OM S , banana baits (see & TH A ON 13 for the details of this experiment) as part of a demographic study of this and a �earch was made for their natural food sources. At this time many trees forming the forest canopy had i04 REVISTA bE IliOLOGIA 11i.DpicAi

already lost their leaves, the ground was hard and the leaf litter very dry. Observations on adult behavior at food sources included evidence, or lack thereof, for (a) agonistic interactions among adults of the same and opposite sex, (b) agonistic interactions of MorPho with other butterflies on the food, (c) courtship in Morpho, and (d) communal feeding. The Barranca forest was revisited during the following December (1973) for the purpose of exploring the feeding behavior of Morpho during the initial stages of the dry season; these Üib servations could then be contrasted with those from the previous January. From December 12 through December 25, 1973, feeding behavior was observed for a total of seven days. At this time the canopy was still fu11 and the forest floor was just beginning to become hard. The last rains had ceased about five weeks prior to these coservations, a date which corresponds very nicely to the cessation of rainfall the previous year. The same type of field observations were made during this second period of study. At this time these butterflies fed exclusively on tree sap associated with tiny wounds on exposed roots of a large canopy tree (Fig. 2), Samanea saman Merrill (Leguminosae) . Despite extensive observation in the forest during the previous January, there were no instances of Morpho or other fruit-eating butterflies feeding on tree sap wounds at that time. Large areas of the forest had been canvassed for this purpose and 6nly one natural food of the butterflies was found (the decayed fruit of the tree Guazuma ulmifolia Lam. - Sterculiaceae). With the discovery of Morpho· on tree sap wounds of S. saman, observations were limited to this "feeding site" in the forest; the observations on each of seven days were usually at 8:00 AM, 10:00 AM, and again between 1:00 and 3 :00 PM. At each time, the sap wounds were approached very quietly to study diurnal patterns of feeding and a count of the morphoes of each sex was made; it is easy to distinguish sexes as the female possesses a series of white dots on the ventral sides of the forewings. During two of the days of study (December 24-25, 1973) the feeding behavior of Morpho was photto. graphed with minimal disturbance. On one day when observations were not made, the tree sap wounds themselves were examined for exudate and for the presence of other insects. At other times searches were made for other food sources being used by Mev-pho. Previous experience with MorPho allows one to determine the approximate age of adults according to the amount of wing damage. As the butterfly ages, the wings become brittle and they begin to tatter and lose scales. When observing adults feeding during the December study period, it was noted whether they were "fresh", "middle", or "old". The tatters in the wings are easy to see at a distance, and with care and practice, one can approach the feeding butterflies to within a meter for closer scrutiny of sex and age. This method of aging adults is by no means accurate, but it does allow one to distinguish in a reliable manner very young ("fresh") individuals from other age groups. It is thus possible td determine if the population is producing new adults during the period of study. YOUNG: FEEDING BEHAVIOR 01' MORPHO iltJTTERFLIES 105

RESULTS

During the dry season (January) Morpho feeds on the decaying fruits of Guazuma ulmijolia (Fig. 3A). Many of the blackened fruits can be found in the seed shadow of the parent free and upon close inspection, ea eh fruit is covered with a thin film of sticky, sweet-smelling substance. But the odor is noticeable to the observer only when the fruit is held to the no�e. The decaying fruit is present throughout most of the dry season after December and it is likely the only food source of Morpho during this period, but butterflies readily come to experimental baits of rotting, crushed banana. Early in the dry season (December) adult Morpho feeds at tiny sap wounds on exposed roots of Samanea saman, either singly, in twos, or in larger groups (Fig. 3-5) The sweet odor of the wounds is only detectable when the nose is placed against it. If banana baits are placed out in the forest near

this feeding site (about 15 m away) , M.orpho, along with Caligo, comes to these within a few minutes. But if such a bait is placed next to the tree wounds (about 1-3 m away), Morpho invariably goes to the latter and ignores the banana; Caligo is never found 0'11 the tree sap wounds. The satyrid . butterfly luna is often found on the tree sap wounds but seldom on the banana (Fig. 4A). Other members of the fruit-feeding ensemble of nymphalid butterflies present in this forest are not found on tree sap wounds but aH of them readily come to banana baits. Pierel'la luna is also found on the fruits of Guazuma but not the others. At least one species of Trigona bee is found daily in large numbers at the sap flows. During December when MorPho feeds 0111 the tree sap wounds of S. saman, the developing fruits of Guazuma are still green and there are few or no fallen fruits on the forest fIoO!. That Morpb.o is actually feeding on G¡¡aztlma fruits, and liquids as­ soeiated with wounds on the big roots of Samanea is indicated by the obser­ vation of the proboseis contacting the substrate (Fig. 5). Furthermore, as feeding continues, there is an accompanying movement of the proboseis in a probing manner for fresh places on the substrate. When the butterflies feed on rotting bananas, there is a noticeable expansion of the abdominal body wall resulting from the prolonged intake of large volumes of liquido In the two natural foods in question, the liquid component consists of a thin sticky film that does not have the watery consistency of juices from the banana bait, and therefore the abdomen does not become distended. Although the tree sap wounds are very small (about 8-10 mm across), several adults of MorPho can feed simultaneously at each one (Fig. 4). The actual wound is a small cleared area where the bark has been scraped away (Fig. 5) and the exposed tissue is sticky to the touch. There are no large accumulations of fungus growth on the wound, although it has been ob�erved elsewhere in wetter Costa Rican sites that MorPho feeds on juices associated . with growth of slimy fungi on trunks of fallen Cordia trees (12). The actual feeding mechanism for taking up nutriénts from sticky substances associated lOó REVISTA bE BIOLOGIA TROPlCÁi.

with tree wounds or fruits is not known for Morpbo, although owing ta the highly viseous condition of the substanees, a liberatian of digestive juices from the proboseis may be involved. Without sueh an adaptation it apparentÍy would be very difficult for these butterflies to exploit sueh viscous foad sources. In

this context, it is again interesting to note that 1 have seen only Morpho and Piereila feeding at these sap flows, perhaps tel the exclusion of many other butterflies that may be able to exploit only more watery substanees. Feeding behavior in MorPbo involves a lack of aggressive interactions among adults when several are present at the same time. The butterflies are spaeed out individualIy on different GuaZtlmd: fruits with little or no. close interaetions. Although conditions are often erowded on tree sap wounds (Fig. 4), there is a notieeable lack of aggression during feeding. In fact, two adults of the same age and sex ean be so close to one another that their Jirobosci actuaHy come into eontact while both are feeding (Fig. 4B ). When another butterfly arrives at a "feeding club", there may be sorne wingflapping of settled indi­ viduals, buí seldom do they cease feeding. Thus "feeding clubs" are ereated by the simultaneous feeding of two or more butterflies at a single sap flow or among two or more sap flows very close tOI one another (i.e., within 2-3 cm) on the same root (Fig. 4) . Both sexes partici­ pate and there are no overt gestures of courtship within the club. 1 have noticed that oceasionally a male wilI attempt to mount a second male by sliding between the closed wings from behind. This behavior usually results in both males settling clown again for feeding. It only occurs when there are severa! adults feeding on one sap flow, and therefore the behavior may be induced by crowding_ Feeding clubs are very temporary associations, in terms of membership and size. They very likely are the result of two factolrs : the necessity for several butterflies to feed together on a type of food supply that is very limited in the forest, and the reinforeement of landing at the same foad patch through visual stimulations of a cruising butterfly by the wing movements of one or mo!re butterflies already on or near that food. The conspicuous blue irideseence of the upper surfaces of the wings function to attract conspecifics in Morpbo. (YoUNG, in preparation), and it may be a powerful means of recruiting individuals to a resouree in forests. lf the feeding club is suddenly disturbed, all of the butterflies take to the wing and flutter in circles over the feeding site; within a few minutes, all of .them settle on nearby understory vegetation and forest floor. The initial intermingling of several morphoes in flight just aboye the feeding site is an explosive event : several shimmering and dazzling-blue, large butterflies "spring up" from the Jorest floor. This behavior is reminiscent of a "predator-eonfusion" syndrome.

Another factor that may be involved in the formation of feeding clubs is the tendency for many butterflies in the population to feed at the same time of day in this habitat. There is greater feeding activity during the early afternoan (13) on Guazuma fruits and erushed banana during ]anuary, and on Samanea sap flows in December (Table 1). Thus the adults do not feed throughout the day, but rather eoncentrate their activity into rather discrete periods. The feeding 1 1 clubs are largest between :00 and 3:00 PH (Tab'e r: and they contain both FEEDlNG "BEHAVIOR OF MORPHO BUTTERFLIES YOUNG: 107

males and females, ,Majes are usually more abundant than females at feeding sites, despite a �ex ratio of unity witnessed in large groups of eggs (12) . the difference is clearly behavioral, with females being more secretive in their habits than males. Oviposition tests with captured females taken from feeding clubs (a total of 5 were tested for fertile eggs-see YOUNG, 11 for the technique) indicate that both mated and unmated females are found in these aggregations.

TABLE 1

Feeding ellsembles of (ld¡¡lt on a sap wound oi an exposed r()ol of the " Morpho peleides tree Scmanea saman (Leguminosae) t e "Barranca site" Í/¡ h lorest, Guanacaste, Costa Rica*

Total' "Fresh" "Middle" "Old" Date Time CfCfó' ó' CfCfó' ó' Cf Cf ó' Cf Cf ó'

Dcc. 12, 1973 8:00 AM o o o o o 2

10:00 AM o 2 o o o 3 o

1:00 PM 2 o 3 2 o o 2

Dcc. 13, 1973 8:00 AM o 2 o o o o o 2

10:00 AM o 4 3 o 8

Dec. 16, 1973 1:30 PM o o o 5 4

Dec. 17, 1973 8 :00 AM o 2 o o o o 3

Dec. 22, 1973 1:00 PM 1 o o o o

2:00 I'M 4 o 2 6

Dec. 2 3, 1973 1:00 PM 5 o o 2 6

2:30" PM o 3 o o 4

Dcc. 24, 1973 1:00 PM 2 2 o 2 6

2:00 PM o 7 6 o 14

3:00 PM 9 7 o 3 2 19

Dec. 2 5, 197 3 1:00 PM o 3 o o 4 (*) Fig. 5.

In faet, one female was so fresh in appea.rance that she must have been only an hour or so old. It is unlikely that the feeding club represents a social unit that enhances the likelihood of successful mating within the-immediate vicinity of the food. Feeding fema!es are very unreseptive and remain very still, with their wings kept tightly shut. Wing flapping in the sexually-dimorphic M. pe/eides DE 108 REVISTA BIOlOGIA TROPICAL

is an important signal given by females during the initia! stage of courtship behavior in cages (Malcolm Barcant, pers. comm.). Feeding clubs are composed of adults of different ages and during December there is a predominance of fresh adults (Table 1); the numbers of old adults are low. This trend in age-structure, also seen later in the dry season, has been ascribed to increased vulnerability of these weakened butterflies due to predation by small vertebrates and to increased mechanical difficulties in flight (13). The frequency olÍ formation of feeding clubs during December may be related to the number of feeding sites available. One aspect of this question is the attractiveness of sap flows on different tree species to the butterflies. Very few data are available on this, but from pre1iminary examination, it is unlikely that many species are preferred. In fact, feeding may be limited to a few leguminous genera in addition tOl Samanea. Another consideration is that the sap flows may originate from the biological activities of other on the forest floor. Small sap wounds on forest trees can be the result of a variety of environmental agents, including abrasion, scraping activities of birds and mammals, and biting activities of insects. Several different ants visit the sap . wounds along with M01Pho; there are three abundant smaIl species : Paratrecbina (Nylanderia) sp., Cyphomyrmes sp., (rimosus group ) , and Pheidole sp. In addition, the larger ant Dolichoderus (Monácis) bispi110SUS (Olivier) is also present in smaIl numbers. The intriguing possibility exists that the sap wounds visited by lVI'orpho are actually "ant tracings", produced by sorne or all of these species by the scraping away of thin layers of bark in search of carbohydrates. lE this is tme, then the feeding clubs of Morpho are created by the foraging activities of the ants. lE the sap flows are ant tracings, then this food type may o'nly be available at certain times of the year and entire1y absent at others, depending upon �easonal changes in foraging activities of the ants. The same is true if the sap flows are oroduced by sorne other animals (birds), and only secondarily visited by ants. SmaIl creepers peck at the large exposed roots of forest trees for a variety of reasons and may create the wounds. Owing to their size and appearance, however, it is very likely that the sap flows are produced by ants (or birds) and not from mechanical abras ion of branches, litter, etc. The sap flows provide a food source for many Morpbo butterflies over a two-week period (Table 1) and probably for a considerably lctnger time. Undoubtedly the butterflies receive carbohydrates from the sapo It is inter­ esting that during the previous ]anuary we did not see any instances of MOl'pho feeding on sap flows, despite intensive fie1d work in that part of the forest where this behavior was to become so evident the foIlowing December.

DISCUSSION

In seasonal tropical environments, food sources may change for animals over relative1y short periods of time and necessitate changes in feeding behavior. Natural selection would favor such i'adaptive shifts" of behavior since animals FEEDING BEHAVIOR OF MORPHO BUTTERFUES YOUNG: 109

that fail to change could suffer a reduction in fitness from severa! sources including (a) outright mortality and massive 1055 of genotypes, (b) reduction in mating success due to weakness from reduction in food, ·and (c) reduction in egg production by the average female in the population. Life span may also decline fo! the average adult, perhaps reducíng the amount of time available for courtship and egg-Iaying. AH of these factors are important for many organ­ isms. In relatively non-seasonal, wet-tropical environments, les s food stress among adults occurs since moist, readily accessible nutrients are present through­ out the year (i.e., slimy fungal growths, decaying fleshy fruits, carrion, etc) . The nutrients in such foods are probably easier to imbibe than those in the viscous foods of Morpho in the Barranca foresto Under less stressful conditions, which for Morpho have been shown to affect population size (10, 13), larger populations are expected. In a very season al environment, larger adult popu­ lations are predicated for the wetter months if more accessible foods are available, and if the butterflies make a shift in feeding behavior. The present findings suggest that even within a period of stress (dry season), there occur shifts in feeding behavior in response to temporal changes in potentia! food. The butterflies do respond to changes in food-types, revealed by the very fast appearance of the insects on the fleshy, rotting bananas placed in the forest, and also by the transition from Samanea sap flows to fallen Guazuma fruits as the dry season advances. In fact, the very fast discovery of experimentally-placed baits by MorPho and other butterflies is indicative of a str<:ssful environment. Assuming that the banana odor does not block the sensory perception of other foods, the quick appearance of these butterflies on the bait and the fact that the same individuals often return to a particular bait day after day (13) indicate that other (natural) foods are in limited supply. The observations of Morpho and Pierella feeding on the Samanea sap flows, even when banana baits are placed nearby, indicate that the highly odoriferous properties of the banana do not necessarily block sensory perception of other foods. Even when the butterflies are baited near fallen Guazuma fruits, many individuals feed on the natural food (13). An intriguing extension of these observations is the possibility that Morpho and others can eventually shift from a viscous natural food to a more watery experimentalIy-placed food such as ba­ nanas. If bananas are easier to imbibe and if they contain the same or similar nutrients needed by ¡1¡forPho as found in natural foods, such ability could be adaptive. In fact, such a shift might occur with natural foods if watery foods become abundant during wetter months in the Barranca forest: there may not be a complete changeover of available foods (i.e., sap flows may occur during the wet season as well) but rather the insects switch over in feeding behavior. Adults of Morpho become very abllndant during July and August in the Ba­ rranca forest and they do not feed on Samanea sap flows or Guazuma fruits dudng these wet months. As food becomes more limited in the habitat, both ' in pll:tch size and distribution of the patches, selection should favor the expression , of cooperative or communal feeding behavior so that several individuals c.a11 : fec;:d, s�cceS$fully lUlVISTA BIOLOGIA TROPICAL. 110 DE

at a given patch. This should be the case since the likelihood and amoun,t of energy involved in locating a different foad patch by a foraging individual become very critical factors in the feeding behavior: the likelihood declines and the amount of energy increases. Both conditions favor communal feeding if individuals are to avoid mortality due to starvation during a stressful periodo A bonus of communal feeding behavior is that conditions are established for "predator confusion" syndromes associated with the disbanding of a feeding club by a sud den disturbance. Such behavior is only adaptive if Morpho is palatable to vertebrate predators. If other factors favor the temporary banding together of these butterflies for feeding on a patchy resource, confusion flight behavior increases the chance for escape of individuals when a feeding club is found by a predator. If alternative foad patches are very easy to find with minimal expenditure of energy, selection might favor the expression of agonistic behavior in the adults at any one food patch. This would be a means of reducing intraspecific competition for food when in fact food is abundant as a patchy but accessible resource. But when food patches are few and widely scattered, as presumably the case for Samanea sap wounds, many individuals in the population will feed in a single or few patches and communal feeding will be expressed. Agonistic behavior was not observed at these food sources. The formation of the Samanea sap wounds used as feeding sites by MorPho and the relationship of the numerous ant species to them remains unresolved. On New Guinea, several species of Paratrechina nest on the floor of the tropical rain forest and workers of at least one species, P. longicornis, abundant in many tropical regions, feed at sugar baits (WILSON, 14). Very little is known about the biology of Cyphomyrmex although the primitive posie tion of this within the Attini is of considerable interest owing to the fungus-culturing habits of a more advanced genus, Atta (14). One wonders if minute fungal growths associated with the the sap flows are the food source of these ants. At least one species of Monacis (1M. debi/is) coUects honeydew from membracids and the presence of M. bispinosus at sap flows may indica te foraging for sugars. The feeding habits of Pheidole are diverse (17). AH of these observations suggest that the ants present at Samanea sap flows are feeding, but the question of whether or not sorne or aH of these ants actuaUy create the feeding sites ("ant tracings") is unresolved. Feeding occurs at various hours of the day, but the largest feeding clubs at Samanea trees are during the mid-afternoon. Especially during the dry months, the hottest hours of the day are in the earIy and middle afternoon, and feeding can result from several selection pressures including feeding at hottest times to avoid desiccation associated with flying, postponement of feeding until after other activities (courtship, oviposition) have taken place, and feeding during hours when forest floor predators are least active. No data are available to support any of these hypotheses, which are not mutualIy exclusive. The most attractive explanation is probably a combination of the first two. Both Morpho' and Caligo are large, dark-colored insects ahd perhaps heat up very qtiick1y in sunshine. For Caligo the problem may ·be more acute since both sides of the YOUNG: FEEDING BEHAVIOR OF MORPHO BUTTERPlIES l11

wings are very dark and generally non-reflective, thus facilitating the uptake of heat. In MorPho the ventral sides of the wings are non-reflective and dark while the upper sides are dark bllt reflective for the most parto Caligo shows a very pronounced bimodal feeding pattern in the Barranca forest, with peaks occurring before 8 :00 AM, and after 4:30 PM during the dry season. This diurnal pattern of adult feeding behavior in Caligo may therefore be a more extreme adaptation for avoiding desiccation and over-heating from large heat uptake dne to the large surface area of the wings. For Morp'ho, with a presumably less thermoregulatory problem, there is but one major peak period of diurnal feding in which the insects become less active during the hottest part of the day. Feeding with the wings closed in the shaded lInderstory of the forest allows them to escape high temperatures. The apparent lack of Morpho feeding on sap flows later in the dry season could either be due to an outright decline in the number of these food patches or increased desiccation rates of these patches, making them very temporary resources for insects. As the dry season advances, wind penetrates deeper into the forest and desiccation rates on a variety of substrates are expected to increase. Nutrient solutions associated with crevices of more textured substrates such as' fallen Guazuma fruits retain this moisture longer than smooth surfaces, thus providing a more permanent food source for insects at the height of the dry season. Clearly the feeding niche of Morpho in the confined Barranca forest includes the exploitation of viscolls films of nlltrients on at least one fruit and one type of tree sap, and there is a succession within weeks from on� resource to the other. There is a shift in food that involves a change in feeding behavior from solitary to commllnal associations at food patches. It is likely that both foods are not the most efficient (in terms of the ability of butterflies to eat them) fo! butterflies such as Morpho: adults wilI feed on crushed bananas when brought into the laboratory and offered this food immediately after a feeding spell on G¡IaZ1t111a fruits or Smnanea sap flows. But individuals that have fed on banana in the wild will not feed immediately on this same fook! in the laboratory, suggesting that less nlltrients are ingested per unit time (along with water) from viscous foods than fcom bananas, so that in fact the dry season foods of Morpho are "stress" foads; during the rainy season there are more watery resources for these and other insects. However, there are two additional interpretations of these observations that must be examined before confirming this conclusion : the butterflies may habituate to a particular food item and will restUne feeding only if a different foad item is offered; and they may cease feeding when their abdomens become sufficiently distended. It is also possible, that despite the fact that banana ju/ce causes rapid abdominal distension and subsequent cessation of feeding, this food item may, in fact, contain less nutrients than natural food items. Banana juice may be mostly water, and analyses are needed Oil its nutritional contents compared to that of Guazuma fruits. 1 believe these findings on the feeding behavior of Morpho during the 112 REVISTA DE BIOLOGlA TROPICAL

tropical dry season have furthet implications for yet another aspect . of the behavioral bioloJgy of neotropical nymphalid butterflies : the behavioral parteras for feeding described bear on the question of phylogenetic relationships within the Morphinae and between this subfami1y and the Satyrinae. This problem has been discussed for the relation of Morpho with the satyrids and (VANE-WruGHT, 9). It is argued that lvlorpho and Antirrhea may be very closely related owing to similarities in adult scent organs, and in the shape of the egg. Vane-Wright (pers. comm. ) has aIso argued that Caligo (in the Morphinae) is not closely related to Morpho owing to departures in egg shape

and adult scent organs associated with courtship behavior. It is therefore very interesting . to note the apparent similarity in foad preferences between Morpho and Pierella in this study, and the absence of Caligo on both of the viscous foods under observation. The similarity of adult . feeding behavior between MorPho and Pieretla (along with other charaeters) presumes a close phylogenetic relationship o!f Morphinae to Satyrinae. The ability of lvlorpho and Prerella to feed on these foods to the apparent exclusion of Caligo and perhaps others provides further support of Vane-Wright's view with different data from an­ other satyrid genus. This behavioral divergence involves morphological and phy­ siological adaptations for exploiting a highly viscous foad; all of the butterflies can feed on watery foods such as bananas, but apparently only sorne (Morpho and Pierella) have exploited a different type of food. The eco!ogical impli­ cations here are clear : assuming that food is limited, interspecific competition among members of the butterfly ensemble exploiting rotting fruits provides selection pressure for sorne members to enter into a new adaptive zone for feeding. Finally, the observations permit a speculative glimpse at the "feeding strategy" of MorPbo that is useful as a starting point for further behavioral studies of considerable general interest. For example, clearly one of the major forces of natural selection operating on any animal population of a mobilc species is how individuals obtain sufficient food for survival and reproduction when this resource is distributed patchily in space and/or time in the habitat. Individuals must locate food patches and retrieve nutrients from them, and the type of feeding behavior expressed will be determined by how this problem is presented to the species. For Morpho in the Barranca forest, at least two

- cases of this problem are apparent : Case 1 a food resource is distributed spatially as numerous large patches and this resource is only seasonally available. This case is exemplified by fallen Guazuma fruits : Guazum'(� trees are nume­ rous bver á large portion of the Bárranca forest and the crown of each tree is large; many of the fallen fruits stay beneath the parent tree. Such a resource distribution in space favors a highly dispersed butterfly population with mínimal or no communal fs:eding behavior. Case II - a food resource is distributed spatially as a few small piltches; it is r�re and hyperclispersed. An illustration of this case is Samanea : t;his tree is rare, not all individuals will have sap flows, and sap flows may or may not be seasonally available as feeding sites fOf; i{l,�ects. Such a yery spati�lly , patc.hy resource with small patch size favors YOUNG: F1lEDlNG BEHAVIOR; OF . MORPHO BUTTERFLlES

the, �xp!essioll oí communal feeding in adult l\tforpho with an increased like­ lihood fo! considerable interaction among bufterflies in feeding ·chibs. B\ü, increased interactions among butterflies, be they related to courtship, predator­ defense, or play, do not indude aggressive encounters since selection wo\Uld disfavor such behavior up to a certain local population density. lf local popu­ lation density becomes very high, then selection would favor in�reased disper­ sal and behavioral patterns promoting aggregation would dissoive. Clearly the paint at which this would happen is determined by the life history pattero of the butterfly. As trees are continually removed from the Barranca forest as a result of increased ?ctivity by local farmers, there may approach a time when such acti"ity will alter the feeding behavior of Morpho. As Guaztlma trees, for example, become scarce, Case 1 aboye becomes an unlikely food-exploitation pattern, and it may force the MorPho population to adopt the Case II strategy with this food item. If the few Samanea trees are 'removed, this would exc1ude the Case' Ir strategy on that food item for the Morpho population. Clearly, these predictions are examples of how destruction of a tropical forest could alter the feeding strategy of a butterfly population that does not exploit flowers. Un!ess the bu tterflies can changeover to other food items not being reduced in number by agricultural practices such as lumbering, the eventual result would be 10(;11 extinction of the species if emigration was not possible. Thus not only is the destruction of larval host plants an important factor in the survival of butterflies in disturbed tropical forests, but c1early the adult food sources are also determining factors. In fact, for lvf.orpho peleides, whose larvae feed on several legumes in forest habitants (12), the greatest stress for survival is likely to be placed on the adults owing to their dependence on fruits, sap flows, and fungi as maj ar food items. These predictions have implications for establishing a system of experi­ mental baits in a confined forest habitat. If piles of bananas are distributed evenly over the forest during December (when feeding is primarily on Samanea sap flows ) active adults of Alorpho will chance upon patches of this new foad source and the pattern of disper5ion would change from Case II to Case l. 011 If the same experiment i5 done during January (when feeding is Guazf.lma fruits), there will be little or no change from the Case 1 pattcrn. At this time, howevér, the butterflies would probably shift fmm GlIctzlIma fnds to bananas. If it is assumed (as argued in this paper) that morphoes prefer bananas to Guazuma fruits, an experimental study of placing small, dispersed patches of bananas in the Barranca forest during January might induce communal feeding

by individuals feeding separately on natural fruits. Such shifts in feeding behavior wilI alter the pattern of gene flow since breeding takes place cluring the dry season · (13).

ACKNOWLEDODJENTS

This research is a by-product of National Science Foundation grant GB- 33060. Logistic support was provided by the Associated Colleges of the Mid- 114 REVISTA DE BIOLOGIA TROPICAL west with the cooperation of ]. Robert Hunter. Identifícatíon of the plant species was made by Luis ]. Poveda (Miuseo Nacional de Costa Rica) and Dieter C. Wasshausen (Smithsonian). The ants were identified by E. O. Wilson (Harvard University) and the butterflies by Lee D. Miller (AUyn Museum · of Entomolgy) . The cornments of an anonymous reviewer of an earlier draft of this paper were mdst helpful.

LlTERATURE CITED

]ANZEN, D. H. 1. 1967. Synchronization of sexual reproduction of trees within the dry season tn . Evo/ution, 21: 620-637.

2. JANZEN, D. H., & T. W. SCHOENER 1968. Differ ences in insect abundance and diversity between wetter and drier sites during a tropical dry season. Ecology, 49 : 96-110.

3. LEWONTlN, R. C. 1965. Selection in and of populations, p. 298-311. 111 J. A. Muore, (ed). ldetts ill Modem Biology. Natural History New York Press,

4. MAYR, E.

1970. Populations, species, evolution. Belknap of Harvard University, alld Press Cambridge, Mass. 453 pp.

5. ORIANS, G. H.

1969. The number uf bird species in some tropical forests. Eco o , 50: 783-801. l gy

6. OWEN, D. F. 1971. Adaptation to man-made environments and phenology in p,.ecis butterflies. Re�·. Zool. Bot. Afrit-aine.r, 83: 361-373.

7. SElTZ, A

1924. Macrolepidoptera oi the world. vol. V. The Americall Rhopalocert1. A. Kernan Verlag, Stuttgart. 1039 pp.

8. SLOBODKIN, B. 1. 1967. Toward a predictive theory of evolution, 187-205. In R. C. Lewontin, p. (ed.). Papulation Biology and Evolu/ion. S racuse Univ. Press, Syracuse, y New York.

Fig. l. General of the forest habitat where the feeding view behavior of adult Morpho peleides was studied during the dry season. (Barranca forest, Guanacaste Province, Costa Rica) . YOUNG: FEEDING BEHAVIOR OF MORPHO BUTTERFLIES 115 REVISTA BIOLOGIA TROPICAL 116 DE

9. VANE-WRlGHT, R. l. 1972. Pre·courtship activity and a new scent organ in butterflíes. Na/ure, 239: 338-340.

10. YOUNG, A. M. 1973. Studies on comparative ecology and ethology in cdult populations of several species of Morpho butterflies (Lepidoptera : Morphidae) . Slud. Neutrop. Fauna, 8: 17-50.

11. YOUNG, A. M. 1974. The rearing of Morpho pe/eídes on . J. Lep. Soc., 28: 90-99.

12. YOUNG, A. M., & A. MUYSHONDT 1973. Notes on the biology of Morpho pe/eMe¡ in Central America. Caríb. J. Scí., 13: 1-49.

13. YOUNG, A. M., & H. THOMASON ]. 1974. The demography oE a confined population of the butterfly Morpbo pe/eídtl during a tropical dry season. SllId. Neo/rop. Fauna 9: 1-34.

14. WILSON, E. O. 1971. The ímee/ socielíes. Belknap Press of Harvard Univ., Cambridge, Mass. 548 pp.

Fig. 2. The forest tree, Samanea saman (Leguminosae), whose large exposed roots have small sap f10ws used a� feeding sites by Morpho and other insects (December 23, 1973). Feeding adults of Morpho líe still against the leaf Iitter. YOUNG: FEEDING BEHAVIOR OF MORPHO BUTTERFLIES 117 118 REVISTA DE BIOLOGIA TROPICAL

F(g. 3. A. Decaying fruits of the tree Guazuma ulmifolia (Ster­ cuJiacae) that are used as a food sourcel by adult Morpho in the Barranca forest during the late dry season. B. fresh male feeding at a sap flow on a root ol Jamanea Jaman (December 24, 1973). C. two fresch males feeding together at a Samanea sap f10w (De- cember 24, 1973). YOUNG: FEEDING BEHAVIOR OF MORPHO BUTTERFLIES 119

3 120 REYISTA DE BIOLOGIA TROPICAL

Fig. 4. A. A feeding club of five morphoes in addition to a single individual of the satyrid Pierella luna (upper right) on two sap flows on a root of Samanea (De­

cember 25, 1973). B. A clQse-up view of two adults of Morpho feeding at a single sap flow (same date). The white area beneath the legs of the individual on the right is the food. Note the very close pro�imity of these individuals and the positions or their anten­ nules. FEEDING BEHAVIOR OF MORPHO BUTT ERFLIES YOUNG: 121

4 122 REVISTA DE BIOLOGIA TROPICAL

Fig. 5 A. Note the proboseis of thlS feeding individual: it is touching the left si de of the small white area (feeding

site) on the root. B. close-up view of feeding: note the proboseis. C. the probaseis of this individual is on the sap flow (white area ). (December 22. 1973 ). YOUNG: FEEDlNG BEHAVIOR OF MORPHO BUTTERFLIES 123

s